Software Bug Location Algorithm on the basis of the Integration of Fuzzy Multi-objective Model
Download as PDF
Bug location is an important part of the software development process. Making full use of the structural features and behavioral features of the program is conducive to improving the bug location efficiency. In this paper, a kind of bug location framework on the basis of the integration of the fuzzy multi-objective model is put forward to carry out the bug location at the class method level on the new version of program during the evolution of software. Firstly, a set of indexes for the measurement of the structural features and the behavioral features are designed. Through the static analysis and the testing program, the feature datasets of the old versions of programs are collected and constructed. At the same time, the old versions of bug information are obtained from the bug tracking system. Secondly, on the basis of the resulting feature datasets and bug information, the univariate analysis is used to filter out the indexes from the measurement indexes that are significantly related to the bugs. Subsequently, the significant indexes that have been selected are used to carry out the multivariate analysis and perform training on the integrated fuzzy multi-objective model. Finally, the feature data set of the new version of program is collected and constructed on the basis of the selected significant indexes. And the fuzzy multi-objective model obtained is used to predict the error probability of each class method. The class method is further inspected in accordance with the error probability in a descending order to achieve the error location. The empirical study of the bug location is carried out on the basis of a set of open source programs. And the results show that the efficiency of bug location can be improved through the integration of the fuzzy multi-objective model.
Bug Location, Fuzzy Multi-objective Model Analysis, Software Measurement, Software Testing